Odometer Calibration

[bramfrank]

GPS is NOT an accurate mechanism for measuring accumulated distance, though it IS accurate for measuring instantaneous speed, once per second.

And how did you arrive at that conclusion?

GPS can be a very accurate mechanism for measuring accumulated distance.

The critical nature being in areas with clear and continuous visibility to multiple satellites. Would I expect to measure distances in downtown Chicago or the twisties and canopy of the forest......no as drop is going to be a problem. If I'm wandering the wide open vistas of Eastern Montana where you can rack up significant digits with a continuous signal....nuts on.

Unless you travel in a straight line (in which case the distance between any two points is a straight line that travels through all of the intervening points) you will not have an accurate reading. And even then it is subject to the instantaneous errors that occur because of atmospherics and multipath. Riding through forests reduces signal strength, but as long as it is above the decode threshold and as long as you have four or more satellites solidly in view it doesn't make a whit of difference.

The major effect is called 'aliasing' and in GPS distance-based measurements it is the shortest distance between sample points. In a straight line the effect is minimized, but is limited to the accuracy of the two points you use for reference and typically when in motion you see +/-5 meter (15 foot) accuracy so that 95% of the time you can be out by as much as 15 feet (for a 30 foot circle) in any direction and up to 5% of the time you can be wildly off.

The GPS-based on-board odometer is based on readings taken once per second and if you happen to turn a corner between samples the system will cut the corner, shaving distance from the total. If you get an anomalous reading (the spec says as much as 5% of the time) THAT will be added to your odometer reading and if the normal errors are such that you are determined to be weaving 30 feet to the left and then 30 feet to the right you will also add distance to the reading.

A very graphic example of the instantaneous error is to watch your Nuvi or Zumo while it is sitting still. My units will easily rack up about 3 kilometers of 'travel' in an hour of sitting on my driveway with a lot of satellites in view. That same error shows up when in motion as well, but you don't notice it partly because the navigator has a function called lock to road that limits displays to the underlying roadway. The 276 and it's cousins can disable the lock to road - it gets pretty ugly. Note that the 2600 series, Streetpilots and 276 use filters to eliminate the additive error effect when they are not in motion.

Tracklogs are worse, for they are not the sum of all of the samples, but simply those samples that the unit feels represent significant change.

Yes, you can use GPS for survey work and you can accurately determine the distance between two points but survey grade equipment has centimeter resolution and is way beyond what we stick on our machines.

 

[Ignacio]

Still then doesn't seem to be a universal statement that "GPS is NOT an accurate mechanism for measuring accumulated distance".

If you know and account for these and other variables they can be a great tool for calibration.

I believe sample rates on the GPS units I've used (Garmin 27xxx series) are faster than once per second. Still can be a factor to be sure if you're riding a twisted path.

Second, the algorithm these units include known roadways as you mentioned with road lock. If you're in an area with only one road nearby and ride that road...it seems to accurately predict you're on the road as opposed to randomly within a 30 foot circle of it. Again, a calibration run that goes from downtown Ely, NV to a point 25 miles away on the only highway out of town and back....tends to be very accurate for measuring accumulated distance. Trying to follow an urban postal route...probably not so accurate.

And I would agree sitting still can produce some weird shifts. That why one should reset the GPS and start riding....and not sit still for a calibration run.

 

[bramfrank]

Nope.

The receivers used in consumer products produce one report per second - and 'lock to road' only impacts on the display, not the odometer or speedometer. It isn't 'locked to the road' in terms of speed or distance - just play back the tracklog.

It's close, and you can often convince a Judge that it is accurate. But statistically it isn't good 'enough'.

 

[Ignacio]

Nope.
The receivers used in consumer products produce one report per second

Hmmm.....Guess I can't find the documentation to verify or disprove that one...I'll have to check the rate when I use my GPS next. I thought it was faster, but I could be wrong.

Did some more thinking about this one and the error you cite of 15 feet is extreme these days. With WAAS available and enabled real-world accuracy is within 1 meter 95% of the time. And I know whenever I've run calibration runs WAAS has been enabled.

So a little 25 mile simulation using good ol' Pythagoras, going 88 feet per second, and a sample every second as you described. That's 1500 samples. In 5 simulations I found an accumulated error of 29.3, 30.7, 31.4 30.8, and 30.7 feet. Pretty close to 30 feet and since the GPS only displays to tenths of a mile anyway (528 feet), and that's all we calibrate to in rallies...I think we're safe to keep using GPS as a reasonably accurate tool for odometer calibration purposes.

 

[Capt. Bob]

I've been criticized for using a Rolatape because of innacuracies that may be caused by temperature variations and road-surface irregularities. I've used a Rolatape for countless traffic accident investigations and reconstructions. I've successfully defended my measurements in court, so have reasonable confidence in them. I'm aware of the precautions to take when using a Rolatape. The road surface was new, smooth and clean asphalt. I walked very slowly, to minimize, or eliminate as much as possible, any "skipping" of the wheel. I checked the Rolatape against two other measuring devices and found it to exactly agree with them.
Thanks, everyone, for sharing your thoughts and offering your opinion.

Comrade,

The criticism was not meant to be directed at you but the device itself. As an old surveyor, I'm use to measuring with a steel tape, tension devices and thermometers, all held in a straight line (as much as possible) with a theodolite. Later we made the switch to lasers and the whole business changed. As I noted before, wheels have their place. Contractors have used them for years and obviously your discipline uses them too but for accurate measurements of land or construction layout (think roads, bridges and buildings) surveyors don't.

Just remember, it's not how many miles per gallon you get but what you pay per gallon.

If that were not true, we would all be riding Vespas or bicycles!

 

[luvtoride]

Each time I get on my FJR I know in my heart that the odometer is wrong. That feeling is always there, it's like a burning coal eating away at my soul, I know that my mileage is wrong.

Oh Anal.... uh, I mean ALAN, you funnyman! So, are you going to cry or something?

 

[ionbeam]

Each time I get on my FJR I know in my heart that the odometer is wrong. That feeling is always there, it's like a burning coal eating away at my soul, I know that my mileage is wrong.

...you funnyman! So, are you going to cry or something?

Just puttin' words in Tolstoy's mouth, I don't really givashit. As long as I push the starter button and the pistons go up and down, and the cams go round and round without the valves kissing the pistions :wub: I'm a happy guy